Modulation of the Clonal Composition in Relapsed CLL: A Study Based on Targeted Deep-Sequencing of ATM, BIRC3, NOTCH1, POT1, SF3B1, SAMHD1 and TP53

Background: Relapse or refractory disease is a challenging clinical problem in the majority of chronic lymphocytic leukemia (CLL) patients. Treatment influences the clonal composition by selection and eventually induction of additional genetic abnormalities.

Aim: To characterize the clonal evolution in relapsed CLL patients by deep-sequencing analysis of mutations in ATM, BIRC3, NOTCH1, POT1, SF3B1, SAMHD1 and TP53.

Patients and Methods: Sequential samples of 20 relapsed CLL patients at three time-points were evaluated: A: at diagnosis (n=16) or in untreated state (n=4), B: at first relapse (n=20) and C: at second relapse (n=2). Patients were treated with diverse treatment schemes and had temporarily achieved either complete or partial remission during the course of the disease. The median time from diagnosis to first-line treatment was 13 months (1 - 169 months). All genes were sequenced by a deep sequencing approach (Illumina, San Diego, CA). IGHV mutational status was determined by direct Sanger sequencing at time-point A. Chromosome banding analysis (CBA) and FISH data on del(17p), del(11q), trisomy 12 (+12), and del(13q) were available in 33/42 and 36/42 samples, respectively.

Results: Initially, samples at first relapse were sequenced. Mutations in SF3B1 (6/20, 30%), TP53 (5/20, 25%), ATM (5/20, 25%), NOTCH1 (4/20, 20%), and SAMHD1 (3/20, 15%) were detected at high frequencies. No mutations were detected in BIRC3 and POT1. In total, 75% of cases presented with at least one mutation (Figure 1): 8 (40%) cases had one, 6 (30%) cases had two and one patient had three genes concomitantly mutated (mut). Patients were also analyzed for IGHV mutational status at diagnosis and presented with unmutated status at a frequency of 85% (17/20).

Subsequently, samples from cases with mutations were analyzed at time-point A. In 12/15 (80%) cases the mutations at relapse were already detectable at time-point A with a similar load indicating presence of the main clone before and after chemotherapy. However, in 7/15 (47%) patients new gene mutations were acquired either additionally to existing mutations (n=4) or in previously wild-type cases (n=3). In 5/7 (71%) cases mutations were located in TP53. TP53mut were the only mutations that were not detected in samples before treatment (sensitivity of 3%). Thus, TP53 mutations might have been initiated by chemotherapy or exist in a minor subclone subsequently selected by chemotherapy. Furthermore, only 4 cases had low-level mutations (3-6% mutation load) at diagnosis in either SAMHD1 or SF3B1 that eventually increased in their burden during disease course.

Of note, in two patients a multibranching clonal evolution could be identified (#2, #9). For patient #2 three time-points were analyzed. At diagnosis 2 ATM mutations were detected with mutation loads of about 20%, each. In the course of the disease these mutations were lost, whereas SF3B1mut showed a stable mutation load in all three time-points of about 40%. In contrast, mutation load of SAMHD1 increased over time from 4% to 87%. CBA was performed at diagnosis and detected independent clones with del(11q) and del(13q). Accordingly, del(11q) detected by FISH at diagnosis was lost and the percentage of cells with del(13q) increased from diagnosis to time-point C. Therefore, patient #2 shows different genetic subclones in parallel that were eradicated or selected by chemotherapy. In patient #9 two SF3B1 mutations were initially detected with the same mutation load of 10%. After treatment one mutation was lost, whereas the load of the second mutation increased indicating at least two different subclones with only one of them being sensitive to chemotherapy. This might be due to different additional aberrations. Indeed, CBA identified two clones: one with +12 alone and one in combination with del(13q). FISH revealed unchanged percentage of +12 at time-point B, whereas del(13q) positive cells were diminished.

Conclusions: In 75% of relapsed CLL cases mutations in SF3B1, TP53, ATM, NOTCH1, and SAMHD1 are present at high frequencies. 80% of these mutations are already detectable before treatment initiation representing the main clone. Remarkably, TP53 mutations were the only mutations that were not detected before but only after chemotherapy.

Figure 1: Distribution of gene mutations in 15 CLL cases with mutations at diagnosis or before treatment (D) and at relapse (R). Red = mutated, grey = wild-type, white = not analyzed.